多地质因素的勘探目标优选——人工神经网络法与多元回归分析法比较研究

APPLICATION OF ARTIFICIAL NEURAL NETWORK AND MULTIPLE REGRESSION ANALYSIS TO OPTIMIZATION OF EXPLORATION PROSPECTS

将人工神经网络法及多元回归分析法分别用于优选预测库车坳陷北带圈闭的勘探目标 ,结果发现人工神经网络法远比多元回归分析法优越。其根本原因是圈闭的优劣与其相关地质因素之间存在着一个复杂的非线性关系 ,人工神经网络法所描述的多因素关系恰是非线性的 ,而多元回归分析法只能描述线性关系。因此 ,当描述多个地质因素的复杂关系时 ,应提倡采用人工神经网络法。当然 ,多元回归分析法也具有人工神经网络所不具备的计算速度快、能较好地表达圈闭优劣与其相关地质因素之间亲疏关系的优点 。

The artificial neural network method and the multiple regression analysis were respectively applied to the optimal prediction of traps in the northern area of Kuga depression.The results show that the former is superior to the latter.There is a complicated nonlinear relationship between trap quality and its related geological factors.The artificial neural network method can describe nonlinear relationship of the multiple geological factors,while the multiple regression analysis method only describes the linear relationship.Hence,it is suggested that the artificial neural network method should be adopted when a complex relationship of multiple geological factors is taken into account.However,the multiple regression analysis method can work as an auxiliary way,because it is in a good position for calculation at a high speed and can express the affinity order between the prospects and its related geological factors,but the artificial neural network method can't do that.